US8797426B2 - Method of image noise reduction - Google Patents

Method of image noise reduction Download PDF

Info

Publication number
US8797426B2
US8797426B2 US13/596,171 US201213596171A US8797426B2 US 8797426 B2 US8797426 B2 US 8797426B2 US 201213596171 A US201213596171 A US 201213596171A US 8797426 B2 US8797426 B2 US 8797426B2
Authority
US
United States
Prior art keywords
frame
video
frames
pixel
value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related, expires
Application number
US13/596,171
Other versions
US20130057733A1 (en
Inventor
Aleksey Anatol'evich BORISOV
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
ZAKRYTOE AKCIONERNOE OBSHCHESTVO IMPUL'S
Zakrytoe akcionernoe obshchestvo "Impul's"
Original Assignee
ZAKRYTOE AKCIONERNOE OBSHCHESTVO IMPUL'S
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by ZAKRYTOE AKCIONERNOE OBSHCHESTVO IMPUL'S filed Critical ZAKRYTOE AKCIONERNOE OBSHCHESTVO IMPUL'S
Assigned to ZAKRYTOE AKCIONERNOE OBSHCHESTVO "IMPUL'S" reassignment ZAKRYTOE AKCIONERNOE OBSHCHESTVO "IMPUL'S" ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: BORISOV, ALEKSEY ANATOL'EVICH
Publication of US20130057733A1 publication Critical patent/US20130057733A1/en
Application granted granted Critical
Publication of US8797426B2 publication Critical patent/US8797426B2/en
Expired - Fee Related legal-status Critical Current
Adjusted expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/80—Camera processing pipelines; Components thereof
    • H04N23/81—Camera processing pipelines; Components thereof for suppressing or minimising disturbance in the image signal generation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N5/00—Details of television systems
    • H04N5/14—Picture signal circuitry for video frequency region
    • H04N5/21—Circuitry for suppressing or minimising disturbance, e.g. moiré or halo
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20024—Filtering details
    • G06T2207/20032—Median filtering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20172—Image enhancement details
    • G06T2207/20182—Noise reduction or smoothing in the temporal domain; Spatio-temporal filtering

Definitions

  • the present method refers to digital image processing namely, to systems processing images obtained with the use of TV camera and intended for image noise reduction.
  • the method of noise reduction in an image is known from the patent EP 0289152, comprising a generation of an image out of the frame video flow.
  • Each video frame is formed from the frame video flow by the following way.
  • a comparison of each frame under processing including appropriate elements of previous frames is performed.
  • the comparison results in defining relations used, considering weight coefficients, to form averaged elements of output frames.
  • the present invention resulted in noise reduction in an image obtained with the use of TV camera.
  • the technical result in the method of noise reduction in an image obtained with the use of TV camera comprising in a video channel a generation of a video flow consisting of groups of frames, having timing interdependency; generation video image from a sequence of output frames obtained by processing the said frame groups by averaging closely adjacent values of appropriate pixels at least in one group of pixels; and the use of averaged values considering weight coefficients to form an output frame is achieved by processing groups of frames comprising odd number, that is 2N+1, where N >1 frames being time-symmetrically juxtaposed against frame under processing with the numbers from ⁇ N to N inclusive, where 0 is the number of the frame under processing, in which the noise is being reduced, 1 is the number of the following frame, ⁇ N is the number of the oldest frame and N is the number of the newest frame; by defining values of frame pixels time-symmetrically juxtaposed against the frame under processing, and their average values; by calculating the absolute value of the said difference with the pixel value of the frame under processing, as a pixel value of the frame under processing are selected considering
  • the method of noise reduction in an image obtained with the use of TV camera comprising a generation of a frame video flow out of frame groups, having timing interdependency; a generation of an image out of output frame series obtained by means of processing of the said frame groups by the use of averaging closely adjacent pixel values in at least one group of frames; and the use of average values considering weight coefficients to form an output frame.
  • the frames are used that time-symmetrically juxtaposed against the frame under processing forth and back in time. It is understood that for calculation of averaged value of any pixel either appropriate pixel values of one frame or other one, or an averaged pixel value of both these frames is used. To select one out of three values the absolute values of the differences of these values with the pixel value of the frame under processing are calculated, then the minimal absolute value of the difference is defined and appropriate pixel value is used for averaging. While averaging selected pixel values are multiplied by coefficients that depend on remoteness extent from the frame under processing. The amount of frames employed in the method and functional coefficient dependence on remoteness extent from the frame under processing define parameters for noise reduction.
  • the method provides noise reduction in an image obtained with the use of TV camera.
  • FIG. 1 shows a processing block diagram
  • FIG. 2 shows a block diagram of the group processing module.
  • noise in a channel can cause errors in an absolute value of the difference calculation and therefore, incorrect operation of the noise reducer so the differences are calculated over three in one row adjacent pixels for a noisy video channel.
  • an absolute value of the difference for the frame under processing and two adjacent pixels are calculated then, a value for further processing by means of a median filter is selected.
  • the claimed method is realized in software as media data provided with a guide to execute the said method.
  • 1 is a group processing module
  • 2 is a multiplier
  • 3 is an adder
  • 4 is a dividing unit.
  • Frame pixels ⁇ 3 ⁇ 3 are divided into three groups each of which incorporates a frame under processing and two time-symmetrically juxtaposed against it.
  • Video—flows from every group are transferred to the module 1 (group processing modules).
  • To every module are also transferred weight coefficients K 1 -K 3 correspondingly.
  • pixels of the frame under processing and coefficient K 0 are transferred to the multiplier 2 .
  • Pixel values are transferred to one adder and, those of coefficients to another one.
  • From the outputs of adders 3 total pixel value and total coefficient value are transferred to the dividing unit 4 . Resulting value of noise reduction is read from the output of the dividing unit divisor 4 .
  • FIG. 2 shows a block diagram of the group processing module 1 where: 5 —is an arithmetical average computing unit, 6 —is an absolute value of the difference computing unit, 7 —is a half absolute value of the difference computing unit, 8 —is absolute value of the difference minimal selection unit.
  • Pixel values of the frame under processing are transferred to the arithmetical average computing unit 5 .
  • Obtained average value and pixel value of the frame under processing are transferred to the half absolute value of the difference computing unit 7 .
  • To absolute value of the difference computing unit 6 are transferred pixel values of symmetrical frames.
  • Three obtained absolute values of the differences are transferred to the absolute value of the difference minimal selection unit 8 . Depending on what absolute value of the difference turns out minimal to the output of unit 8 is transferred either pixel value of one of the symmetrical frames or average value of both these frames.
  • the coefficient K is multiplied by 2.
  • weight coefficient values can be selected in correlation with frame pair remoteness from the frame under processing
  • weight coefficient values can be calculated considering one parameter equal to a width of a bell-shaped curve as a function of coefficient via frame number, that width defines the extent of noise reduction;
  • amount of frames employed for processing is selected to be double width of a bell-shaped curve
  • absolute value of the difference is calculated for a processed pixel and two adjacent pixels in a row and as a resulted value is selected one of three values employing a median filter.
  • a distinctive characteristic of the claimed method is that for noise reduction time-symmetrically juxtaposed with respect to the frame under processing forth and back in time frames are used. It is understood that for calculation of averaged value of any pixel either appropriate pixel values of one frame or other one, or an averaged pixel value of both these frames is used. To select one out of three values the absolute values of the differences of these values with the pixel value of the frame under processing are calculated, then the minimal absolute value of the difference is defined and appropriate pixel value is used for averaging. While averaging selected pixel values are multiplied by coefficients that depend on remoteness extent from the frame under processing. Amount of frames employed in the method and functional coefficient dependence on remoteness extent from the frame under processing define parameters for noise reduction.
  • the absolute values of the differences of these values with the pixel value of the frame under processing are calculated then the minimal absolute value of the difference is defined and appropriate pixel value is used for averaging. While averaging selected pixel values are multiplied by coefficients that depend on remoteness extent from the frame under processing. Amount of frames employed in the method and functional coefficient dependence on remoteness extent from the frame under processing define parameters for noise reduction.
  • the method is carried out as follows:
  • a sequence comprising N frames is obtained, where N is the frame number. Then the odd number of frames are processed simultaneously, that is N+1 frames being time—symmetrically juxtaposed against the frame under processing with the numbers from ⁇ N to N inclusive, where 0 is the number of the current frame, in which the noise is being reduced, ⁇ 1 is the number of the previous frame, 1 is the number of the next frame—N is the number of the oldest frame and N is the number of the newest frame.
  • P-N(x,y) is a pixel value of the oldest frame
  • P-N+1(x,y) is a pixel value of the next frame
  • P 0 ( x,y ) is a pixel value of the frame under processing
  • PN(x,y) is a pixel value of the newest frame.
  • y is an image row number
  • x is a pixel position in the image row.
  • the following algorithm is used to calculate an output frame.
  • Coefficients for frame pairs KM are calculated, where M is a number of a frame pair time—symmetrically juxtaposed with respect to the frame under processing, correspondingly M varies from 1 to N, the sum of products of pixel values and coefficients ⁇ PK is equated to 0, the sum of coefficients ⁇ K is equated to 0, the P 0 ( x,y ) value of the pixel under processing is multiplied by the coefficient K 0 and added to ⁇ PK, K 0 is added to ⁇ K for all frame pairs time—symmetrically juxtaposed with respect to the frame under processing; in other words for all M values from 1 to N, the following operations are implemented:
  • ⁇ PK is added to P-M(x,y)*KM, and to ⁇ K is added KM. If a minimal value turned out to be DM, so to ⁇ PK is added PM(x,y)*KM, and to ⁇ K is also added KM. If a minimal value turned out to be DA, so to ⁇ PK is added (P-M(x,y)+PM(x,y))*KM, and to ⁇ K is added 2*KM.
  • a required extent of noise reduction determines KM coefficient selection.
  • the first set corresponds to maximal extent of noise reduction with all coefficients being equal to 1.
  • the second set corresponds to the complete absence of noise reduction with all coefficients equal to 0, except for K 0 that is equal to 1.
  • the coefficients are calculated such that the coefficients for the frames juxtaposed to the current frame (frame “0”) be close to K 0 in their value and decrease as the frame number increases.
  • Other variants can be developed in different ways.
  • the coefficient value via frame number dependence is given by a bell-shaped curve; the width of the bell-shaped curve will determine the noise reduction extent.
  • the method assumes that there is a delay of N in image output. Therefore, when selecting the number of frames a compromise between desired maximal noise reduction extent and minimal delay in image output shall be taken into consideration.
  • the claimed method can be realized with the use of known hardware.
  • An example of an embodiment for carrying out the method is shown in FIGS. 1 and 2 .

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Picture Signal Circuits (AREA)
  • Image Processing (AREA)
  • Studio Devices (AREA)

Abstract

A method of noise reduction in an image obtained with the use of TV camera comprising a generation of a frame video flow out of frame groups, having timing interdependency; a generation of an image out of output frame series obtained by means of processing of the said frame groups by the use of averaging closely adjacent pixel values in at least one group of frames; and the use of average values considering weight coefficients to form an output frame.

Description

RELATED APPLICATIONS
This application claims priority to Eurasian Patent Application No. EA201101158, filed Sep. 1, 2011, which is incorporated herein by reference in its entirety.
FIELD OF THE INVENTION
The present method refers to digital image processing namely, to systems processing images obtained with the use of TV camera and intended for image noise reduction.
BACKGROUND OF THE INVENTION
Quality of images obtained with the use of TV camera can deteriorate due to noise. This noise is required to be reduced what will result in image quality improving.
The method of noise reduction in an image obtained with the use of TV camera is known from the patent US2009154825, published on 18 Jun. 2009, comprising a generation of an output frame out of the frame video flow, having timing interdependency; averaging closely adjacent values of frame elements in one group of frames; and the use of average values considering weight coefficients to form an output frame.
The method of noise reduction in an image is known from the patent EP 0289152, comprising a generation of an image out of the frame video flow. Each video frame is formed from the frame video flow by the following way. A comparison of each frame under processing including appropriate elements of previous frames is performed. The comparison results in defining relations used, considering weight coefficients, to form averaged elements of output frames.
SUMMARY OF THE INVENTION
The present invention resulted in noise reduction in an image obtained with the use of TV camera.
The technical result in the method of noise reduction in an image obtained with the use of TV camera comprising in a video channel a generation of a video flow consisting of groups of frames, having timing interdependency; generation video image from a sequence of output frames obtained by processing the said frame groups by averaging closely adjacent values of appropriate pixels at least in one group of pixels; and the use of averaged values considering weight coefficients to form an output frame is achieved by processing groups of frames comprising odd number, that is 2N+1, where N >1 frames being time-symmetrically juxtaposed against frame under processing with the numbers from −N to N inclusive, where 0 is the number of the frame under processing, in which the noise is being reduced, 1 is the number of the following frame, −N is the number of the oldest frame and N is the number of the newest frame; by defining values of frame pixels time-symmetrically juxtaposed against the frame under processing, and their average values; by calculating the absolute value of the said difference with the pixel value of the frame under processing, as a pixel value of the frame under processing are selected considering weight coefficients such pixel values that correspond to the minimal absolute value of the difference.
The method of noise reduction in an image obtained with the use of TV camera comprising a generation of a frame video flow out of frame groups, having timing interdependency; a generation of an image out of output frame series obtained by means of processing of the said frame groups by the use of averaging closely adjacent pixel values in at least one group of frames; and the use of average values considering weight coefficients to form an output frame.
For noise reduction the frames are used that time-symmetrically juxtaposed against the frame under processing forth and back in time. It is understood that for calculation of averaged value of any pixel either appropriate pixel values of one frame or other one, or an averaged pixel value of both these frames is used. To select one out of three values the absolute values of the differences of these values with the pixel value of the frame under processing are calculated, then the minimal absolute value of the difference is defined and appropriate pixel value is used for averaging. While averaging selected pixel values are multiplied by coefficients that depend on remoteness extent from the frame under processing. The amount of frames employed in the method and functional coefficient dependence on remoteness extent from the frame under processing define parameters for noise reduction.
The method provides noise reduction in an image obtained with the use of TV camera.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 shows a processing block diagram; and
FIG. 2 shows a block diagram of the group processing module.
 DESCRIPTION OF THE PREFERRED EMBODIMENTS
Since noise in a channel can cause errors in an absolute value of the difference calculation and therefore, incorrect operation of the noise reducer so the differences are calculated over three in one row adjacent pixels for a noisy video channel. For that purpose an absolute value of the difference for the frame under processing and two adjacent pixels are calculated then, a value for further processing by means of a median filter is selected.
The claimed method is realized in software as media data provided with a guide to execute the said method.
FIG. 1 shows a processing block diagram for N=3, where:
1 is a group processing module, 2 is a multiplier, 3 is an adder, 4 is a dividing unit.
Frame pixels −3÷3 are divided into three groups each of which incorporates a frame under processing and two time-symmetrically juxtaposed against it. Video—flows from every group are transferred to the module 1 (group processing modules). To every module are also transferred weight coefficients K1-K3 correspondingly. Besides, pixels of the frame under processing and coefficient K0 are transferred to the multiplier 2. From the outputs of the module 1 pixel and coefficients values which are used to calculate the said pixel values are transferred to the adders 3. Pixel values are transferred to one adder and, those of coefficients to another one. From the outputs of adders 3 total pixel value and total coefficient value are transferred to the dividing unit 4. Resulting value of noise reduction is read from the output of the dividing unit divisor 4.
FIG. 2 shows a block diagram of the group processing module 1 where: 5—is an arithmetical average computing unit, 6—is an absolute value of the difference computing unit, 7—is a half absolute value of the difference computing unit, 8—is absolute value of the difference minimal selection unit.
Four values are transferred to the group processing module: pixel values of the frame under processing, pixel values time-symmetrically juxtaposed with respect to a frame and coefficient by which pixel values of this group are multiplied. Pixel values of symmetrical frames are transferred to the arithmetical average computing unit 5. Obtained average value and pixel value of the frame under processing are transferred to the half absolute value of the difference computing unit 7. To absolute value of the difference computing unit 6 are transferred pixel values of symmetrical frames. Three obtained absolute values of the differences are transferred to the absolute value of the difference minimal selection unit 8. Depending on what absolute value of the difference turns out minimal to the output of unit 8 is transferred either pixel value of one of the symmetrical frames or average value of both these frames. In addition, if an average value is selected in the unit 8, the coefficient K is multiplied by 2.
In specific embodiment versions of the claimed method:
weight coefficient values can be selected in correlation with frame pair remoteness from the frame under processing;
weight coefficient values can be calculated considering one parameter equal to a width of a bell-shaped curve as a function of coefficient via frame number, that width defines the extent of noise reduction;
amount of frames employed for processing as a parameter of noise reduction;
amount of frames employed for processing is selected to be double width of a bell-shaped curve;
absolute value of the difference is calculated for a processed pixel and two adjacent pixels in a row and as a resulted value is selected one of three values employing a median filter.
A distinctive characteristic of the claimed method is that for noise reduction time-symmetrically juxtaposed with respect to the frame under processing forth and back in time frames are used. It is understood that for calculation of averaged value of any pixel either appropriate pixel values of one frame or other one, or an averaged pixel value of both these frames is used. To select one out of three values the absolute values of the differences of these values with the pixel value of the frame under processing are calculated, then the minimal absolute value of the difference is defined and appropriate pixel value is used for averaging. While averaging selected pixel values are multiplied by coefficients that depend on remoteness extent from the frame under processing. Amount of frames employed in the method and functional coefficient dependence on remoteness extent from the frame under processing define parameters for noise reduction.
To select one out of three values the absolute values of the differences of these values with the pixel value of the frame under processing are calculated then the minimal absolute value of the difference is defined and appropriate pixel value is used for averaging. While averaging selected pixel values are multiplied by coefficients that depend on remoteness extent from the frame under processing. Amount of frames employed in the method and functional coefficient dependence on remoteness extent from the frame under processing define parameters for noise reduction.
The method is carried out as follows:
A sequence comprising N frames is obtained, where N is the frame number. Then the odd number of frames are processed simultaneously, that is N+1 frames being time—symmetrically juxtaposed against the frame under processing with the numbers from −N to N inclusive, where 0 is the number of the current frame, in which the noise is being reduced, −1 is the number of the previous frame, 1 is the number of the next frame—N is the number of the oldest frame and N is the number of the newest frame.
For each pixel of the frame under processing a new value is calculated in the following way.
With the aim of clarification let us introduce the following notations:
P-N(x,y) is a pixel value of the oldest frame,
P-N+1(x,y) is a pixel value of the next frame,
P0(x,y) is a pixel value of the frame under processing,
PN(x,y) is a pixel value of the newest frame.
Here:
y is an image row number,
x is a pixel position in the image row.
The following algorithm is used to calculate an output frame.
Coefficients for frame pairs KM are calculated, where M is a number of a frame pair time—symmetrically juxtaposed with respect to the frame under processing, correspondingly M varies from 1 to N, the sum of products of pixel values and coefficients ΣPK is equated to 0, the sum of coefficients ΣK is equated to 0, the P0(x,y) value of the pixel under processing is multiplied by the coefficient K0 and added to ΣPK, K0 is added to ΣK for all frame pairs time—symmetrically juxtaposed with respect to the frame under processing; in other words for all M values from 1 to N, the following operations are implemented:
absolute values of the differences of pixel values firstly, with the frame numbers 0 and then 0 and M; 0 and an average pixel value for frame numbers—M and M are calculated.
D−M=|P0(x,y)−P−M(x,y)|
DM=|P0(x,y)−PM(x,y)|
DA=|P0(x,y)−(P−M(x,y)+PM(x,y))/21/2
the minimal values of D-M, DM, DA are determined
if a minimal value turned out to be D-M, ΣPK is added to P-M(x,y)*KM, and to ΣK is added KM. If a minimal value turned out to be DM, so to ΣPK is added PM(x,y)*KM, and to ΣK is also added KM. If a minimal value turned out to be DA, so to ΣPK is added (P-M(x,y)+PM(x,y))*KM, and to ΣK is added 2*KM.
After processing of all M frame pairs the resulted pixel value can be computed in the following way
P out(x,y)=ΣPK/ΣK.
A required extent of noise reduction determines KM coefficient selection.
There are two special coefficient sets. The first set corresponds to maximal extent of noise reduction with all coefficients being equal to 1. The second set corresponds to the complete absence of noise reduction with all coefficients equal to 0, except for K0 that is equal to 1. For noise reduction of intermediate values the coefficients are calculated such that the coefficients for the frames juxtaposed to the current frame (frame “0”) be close to K0 in their value and decrease as the frame number increases. Other variants can be developed in different ways.
In one embodiment the coefficient value via frame number dependence is given by a bell-shaped curve; the width of the bell-shaped curve will determine the noise reduction extent. The example of such a function:
For i<R
Ki=(cos(i·π/R)+1.0)
For i>R
Ki=0.
Here R is a parameter determining noise reduction extent. At R>>N all coefficients are equal and noise reduction extent is minimal. At R<1 all values except the average are equal to “0”, a noise reducer will be practically OFF. R=2 means that the middle frame will be ON with the coefficient of 0.5, and two adjacent ones—with coefficients of 0.25, that is the width of the bell will be 2.
The method assumes that there is a delay of N in image output. Therefore, when selecting the number of frames a compromise between desired maximal noise reduction extent and minimal delay in image output shall be taken into consideration.
In the claimed method such artifacts as brightness jump tailing are decreased significantly due to their being positioned symmetrically with respect to the middle frame, that is the artifact is allocated on two sides of the middle frame and its visibility diminishes sharply. In addition, when brightness is changed gradually, due to using frames positioned symmetrically with regard to the middle frame, noise reduction extent increases significantly even when the image changes. Besides, if the image stops changing with respect to the middle frame, or does not change up to the current frame the noise reduction extent keeps remaining high. Moreover, in the claimed embodiment setting time is limited by a selected number of cells.
The claimed method can be realized with the use of known hardware. An example of an embodiment for carrying out the method is shown in FIGS. 1 and 2.

Claims (6)

What is claimed is:
1. A method of reducing noise in a video produced using a TV camera, the method comprising
generating in a video channel a video flow with at least one video frame group, the at least one video frame group comprising an odd number of 2N+1 video frames having time interdependence,
wherein N is ≧1, and
wherein the video frame to be processed for reducing noise to produce an output video frame is a middle video frame in the at least one video frame group, and wherein remaining video frames in the at least one video frame group are N pairs of video frames juxtaposed time-symmetrically with respect to the middle video frame;
determining a pixel value of a corresponding pixel positioned in each of the 2N+1 video frames;
calculating for each pair of video frames an average pixel value for corresponding pixels positioned in the video frames of the pair;
calculating for each pair of video frames an absolute value of a difference between the calculated average pixel value and the determined pixel value of a corresponding pixel in the middle frame,
calculating for each video frame of each pair an absolute value of a difference between the determined pixel value of a corresponding pixel positioned in a pair of video frames and the determined pixel value of a corresponding pixel positioned in the middle video frame;
determining a minimal of all said absolute values; and
selecting a pixel value for a corresponding pixel in the output frame as a weighted pixel value corresponding to the determined minimal absolute value.
2. The method of claim 1, wherein weight coefficients selected for the weighted pixel value depend on the distance between the frame and remaining frames in a frame pair.
3. The method of claim 1, wherein the weight coefficients are determined using a parameter equal to a width of a bell-shaped function determining a weight coefficient for a frame number, and wherein the width determines extent of the noise reduction.
4. The method of claim 1, wherein the number of frames in a video channel used for processing is a parameter for the noise reduction.
5. The method of claim 1, wherein the number of frames in a video channel used for processing is twice a width of a bell-shaped curve.
6. The method of claim 1, further comprising calculating an absolute value of a difference for a processed pixel and two pixels adjacent to the processed pixel in a row and selecting a resulting value from the three values using a median filter.
US13/596,171 2011-09-01 2012-08-28 Method of image noise reduction Expired - Fee Related US8797426B2 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
EA201101158 2011-09-01
EA201101158A EA201101158A1 (en) 2011-09-01 2011-09-01 METHOD OF REDUCING NOISE IN VIDEO IMAGE

Publications (2)

Publication Number Publication Date
US20130057733A1 US20130057733A1 (en) 2013-03-07
US8797426B2 true US8797426B2 (en) 2014-08-05

Family

ID=46614786

Family Applications (1)

Application Number Title Priority Date Filing Date
US13/596,171 Expired - Fee Related US8797426B2 (en) 2011-09-01 2012-08-28 Method of image noise reduction

Country Status (6)

Country Link
US (1) US8797426B2 (en)
EP (1) EP2565842A1 (en)
JP (1) JP2013055652A (en)
KR (1) KR20130025351A (en)
CN (1) CN102968773A (en)
EA (1) EA201101158A1 (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103702016B (en) * 2013-12-20 2017-06-09 广东威创视讯科技股份有限公司 Vedio noise reduction method and device
CN105046677B (en) * 2015-08-27 2017-12-08 安徽超远信息技术有限公司 A kind of enhancing treating method and apparatus for traffic video image
EP3466051A1 (en) * 2016-05-25 2019-04-10 GoPro, Inc. Three-dimensional noise reduction
CN110246087B (en) * 2018-03-07 2021-06-04 舜宇光学（浙江）研究院有限公司 System and method for removing image chroma noise by referring to multi-resolution of multiple channels

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0289152B1 (en) 1987-04-30 1993-08-04 Shimadzu Corporation Apparatus for processing an x-ray image
US6037986A (en) * 1996-07-16 2000-03-14 Divicom Inc. Video preprocessing method and apparatus with selective filtering based on motion detection
US6058143A (en) * 1998-02-20 2000-05-02 Thomson Licensing S.A. Motion vector extrapolation for transcoding video sequences
US20020158971A1 (en) 2001-04-26 2002-10-31 Fujitsu Limited Method of reducing flicker noises of X-Y address type solid-state image pickup device
US20030103568A1 (en) 2001-11-30 2003-06-05 Samsung Electronics Co., Ltd. Pixel data selection device for motion compensated interpolation and method thereof
US20080253456A1 (en) 2004-09-16 2008-10-16 Peng Yin Video Codec With Weighted Prediction Utilizing Local Brightness Variation
US20090154825A1 (en) 2007-12-14 2009-06-18 Intel Corporation Reduction filter based on smart neighbor selection and weighting (nrf-snsw)
US20110228167A1 (en) * 2007-07-11 2011-09-22 Hiroshi Sasaki Image processing apparatus, image processing method, and recording medium for image processing program
US20120114041A1 (en) * 2010-11-08 2012-05-10 Canon Kabushiki Kaisha Motion vector generation apparatus, motion vector generation method, and non-transitory computer-readable storage medium

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100672328B1 (en) * 2005-01-18 2007-01-24 엘지전자 주식회사 Apparatus for estimation noise level of video signal
TWI324013B (en) * 2006-02-22 2010-04-21 Huper Lab Co Ltd Video noise reduction method using adaptive spatial and motion-compensation temporal filters

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0289152B1 (en) 1987-04-30 1993-08-04 Shimadzu Corporation Apparatus for processing an x-ray image
US6037986A (en) * 1996-07-16 2000-03-14 Divicom Inc. Video preprocessing method and apparatus with selective filtering based on motion detection
US6058143A (en) * 1998-02-20 2000-05-02 Thomson Licensing S.A. Motion vector extrapolation for transcoding video sequences
RU2251820C2 (en) 1998-02-20 2005-05-10 Томсон Лайсенсинг С.А. Extrapolation of movement vector for video sequence code conversion
US20020158971A1 (en) 2001-04-26 2002-10-31 Fujitsu Limited Method of reducing flicker noises of X-Y address type solid-state image pickup device
US20030103568A1 (en) 2001-11-30 2003-06-05 Samsung Electronics Co., Ltd. Pixel data selection device for motion compensated interpolation and method thereof
US7720150B2 (en) 2001-11-30 2010-05-18 Samsung Electronics Co., Ltd. Pixel data selection device for motion compensated interpolation and method thereof
US20080253456A1 (en) 2004-09-16 2008-10-16 Peng Yin Video Codec With Weighted Prediction Utilizing Local Brightness Variation
US20110228167A1 (en) * 2007-07-11 2011-09-22 Hiroshi Sasaki Image processing apparatus, image processing method, and recording medium for image processing program
US20090154825A1 (en) 2007-12-14 2009-06-18 Intel Corporation Reduction filter based on smart neighbor selection and weighting (nrf-snsw)
US20120114041A1 (en) * 2010-11-08 2012-05-10 Canon Kabushiki Kaisha Motion vector generation apparatus, motion vector generation method, and non-transitory computer-readable storage medium

Also Published As

Publication number Publication date
US20130057733A1 (en) 2013-03-07
JP2013055652A (en) 2013-03-21
EA016695B1 (en) 2012-06-29
EA201101158A1 (en) 2012-06-29
KR20130025351A (en) 2013-03-11
CN102968773A (en) 2013-03-13
EP2565842A1 (en) 2013-03-06

Similar Documents

Publication Publication Date Title
TWI231706B (en) Method and system for advanced edge-adaptive interpolation for interlace-to-progressive conversion
US8797426B2 (en) Method of image noise reduction
US9105081B2 (en) Protection filter for image and video processing
US8792746B2 (en) Image processing apparatus, image processing method, and program
US20090040374A1 (en) Image processing apparatus and method of controlling the same
US8189105B2 (en) Systems and methods of motion and edge adaptive processing including motion compensation features
US8238685B2 (en) Image noise reduction method and image processing apparatus using the same
CN1860779B (en) Image processing apparatus and method
CN105890768B (en) A kind of method and device of Infrared Image Non-uniformity Correction
US20100259675A1 (en) Frame rate conversion apparatus and frame rate conversion method
US20100026888A1 (en) Image processing method and system with repetitive pattern detection
US9542617B2 (en) Image processing device and image processing method for correcting a pixel using a corrected pixel statistical value
KR102411033B1 (en) IPS bias-compensated noise reduction systems and methods
US20120242896A1 (en) Video image processing apparatus and video image processing method
CN104254872A (en) Image processing method and image processing device
CN111294545B (en) Image data interpolation method and device, storage medium and terminal
US20100272372A1 (en) Image Processing Apparatus and Image Processing Method
CN102497523B (en) Median-filtering de-interlacing method
EP2843617B1 (en) Image processing method and image processing device
JP2016080950A (en) Image processing device, method for controlling the same, image display device, and computer program
EP3146706A1 (en) Video contrast enhancement with sub-segments
GB2422264A (en) Spatial interpolation along a slope calculated from image pixel values
CN113261276B (en) De-interlacing interpolation method, de-interlacing interpolation device, de-interlacing interpolation system, video processing method and storage medium
CN103927740A (en) Method and device for obtaining point spread function of image
Park et al. Adaptive weighting scheme for edge-based line interpolation

Legal Events

Date Code Title Description
AS Assignment

Owner name: ZAKRYTOE AKCIONERNOE OBSHCHESTVO "IMPUL'S", RUSSIA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:BORISOV, ALEKSEY ANATOL'EVICH;REEL/FRAME:028858/0670

Effective date: 20120821

FEPP Fee payment procedure

Free format text: MAINTENANCE FEE REMINDER MAILED (ORIGINAL EVENT CODE: REM.)

LAPS Lapse for failure to pay maintenance fees

Free format text: PATENT EXPIRED FOR FAILURE TO PAY MAINTENANCE FEES (ORIGINAL EVENT CODE: EXP.); ENTITY STATUS OF PATENT OWNER: SMALL ENTITY

STCH Information on status: patent discontinuation

Free format text: PATENT EXPIRED DUE TO NONPAYMENT OF MAINTENANCE FEES UNDER 37 CFR 1.362

FP Lapsed due to failure to pay maintenance fee

Effective date: 20180805